# Views (templates) > There is no single dashboard. A **view** is a saved way of asking your AI host to render your portfolio — and because HeadlessTracker just returns structured data, the same numbers can become an overview, a trader screen, a risk audit, or anything else you ask for. Each built-in view ships as an **MCP prompt**: in Claude Desktop (or any MCP host) it shows up as a one-click prompt that tells the model exactly which HeadlessTracker tools to call and how to lay out the result. No new UI, no rendering code — the host draws it. That's the whole thesis: **build the data layer, let the AI host be the renderer.**

Portfolio overview view: net worth across six venues, a donut allocation chart, a by-venue split bar, and per-venue holdings.

_{The portfolio-dashboard view. Sample data — the picture is whatever the host renders on top of the tool output.}

--- ## Built-in views ### 📊 `portfolio-dashboard` — "What do I own across everything?" A complete multi-section dashboard in one artifact: total value, allocation by asset class, top positions, P&L summary, and a Polymarket section. **Calls:** `get_holdings`, `get_allocations` (by asset class + by symbol), `get_pnl`, `get_polymarket_positions`. **Use it:** pick the **Portfolio Dashboard** prompt in your MCP host, or just ask "build me a portfolio dashboard." ### 🗓️ `weekly-review` — "How did I do this week?" A 7-day review: window delta (your current basket valued at 7-day-old prices vs now), biggest movers, trades this week grouped by venue, and one plain-English observation. Honest about what the delta does and doesn't capture. **Calls:** `get_pnl` (`timeframe: 7d`), `get_holdings`, `get_transactions` (`since: 7d`). **Use it:** the **Weekly Review** prompt, or "give me a weekly portfolio review."

Weekly review view: a 7-day window delta headline with an honest caveat, biggest movers (gainers and losers), trades this week grouped by venue, and one plain-English observation.

### 🛡️ `risk-check` — "Is anything structurally risky here?" A concentration and risk audit scored PASS / WARN / ALERT across single-position dominance, venue concentration, stablecoin reserve, prediction-market overweight, and per-chain concentration. **Calls:** `get_holdings`, `get_allocations` (by symbol, by asset class, by connector). **Use it:** the **Risk Check** prompt, or "do a risk check on my portfolio."

Risk check view: a PASS/WARN/ALERT scorebar and a table scoring single-position concentration, venue concentration, stablecoin reserve, prediction-market weight, and per-chain concentration, with a one-line summary.

### 🧩 `diversification-check` — "Am I actually diversified, or just holding 10 things that move together?" · ★ first community view, proposed by [@hermessfo](https://www.moltbook.com) Groups every holding into **correlation clusters** (BTC-beta, ETH & ecosystem, SOL & ecosystem, other alts, stablecoins, prediction) instead of by chain or venue — so the largest cluster's % is your *real* concentration, no matter how many tickers or exchanges it's spread across. Calls out the **multi-venue illusion**: holding correlated assets on five venues spreads custody risk, not market risk. Honest that it's a structural grouping, not a coefficient computed from price history. **Calls:** `get_holdings`, `get_allocations` (by symbol). **Use it:** the **Diversification Check** prompt, or "am I genuinely diversified?"

Diversification check view: a 79% headline showing how much of the book is one correlated crypto-beta bet, a correlated-vs-uncorrelated split bar, a table of six correlation clusters with USD and percentages, and a callout explaining the multi-venue illusion (custody risk vs market risk).

_{The first view proposed by someone other than me. @hermessfo described it on Moltbook; I wired it up and it now ships in the box. This is exactly the loop below — your idea, my implementation, your name on it.}

--- ## Example: a view you could contribute This isn't a built-in — it's the kind of view **you** can propose. Same demo portfolio, asked a different question, rendered as a trader screen:

Trader view of the same portfolio: a 30-day P&L headline with a sparkline, today's movers, and an open-leverage table of Hyperliquid perp positions.

_{A "trader" view — "How are my positions doing?" Same data and tools, a P&L-first render.}

--- ## Contribute a view — _not a connector_ Adding a connector is a lot of work (an API, credentials, tests). **A view is not.** A view is just: a question, the shape of the answer, and which existing tools it calls. If you can describe it, it can ship. **The deal:** propose a view concept → I build it as a real MCP prompt → you're **credited by handle** in this gallery → I reply with a screenshot of *your* view rendered. (Humans and AI agents both welcome — if you live on an agent network, drop the concept there and tag me.) ### How to propose one Open a [**View idea**](https://github.com/tamasPetki/HeadlessTracker/issues/new?template=view_idea.md) issue with: - **The question** it answers (e.g. "what's my stablecoin runway?"). - **The shape**: sections, table vs chart, what to highlight. - **Which existing tools** it would call (see the [tool list](README.md)). A good view reuses what's already there. - A handle you'd like to be credited as (optional). ### What I'll do with it (and the rules) A submitted concept is an **idea**, and the only thing I do with it is design a prompt from it myself — in my own words, calling only existing HeadlessTracker tools, with no external calls, held to the same honesty and "not financial advice" bar as everything else. I don't ship submitted text verbatim and I never run submitted code. That keeps every view safe for the people who install it. You get full credit for the idea; the implementation is reviewed and mine. > A view template is a **prompt**, not a UI — so it fits the "host is the renderer" thesis cleanly. (New UI surfaces still don't; see [CONTRIBUTING.md](CONTRIBUTING.md).) --- _Data aggregation only — not financial advice._